WITHDRAWN: Structural and magnetic properties of the new brownmillerite oxides La1−xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3)

Accepted Manuscript Comment on “structural and magnetic properties of the new brownmillerite oxides La1−xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3)” Josie E. Auckett, Chris D. Ling PII:

S0254-0584(16)30575-2

DOI:

10.1016/j.matchemphys.2016.07.055

Reference:

MAC 19079

To appear in:

Materials Chemistry and Physics

Received Date: 9 March 2016 Revised Date:

18 June 2016

Accepted Date: 20 July 2016

Please cite this article as: J.E. Auckett, C.D. Ling, Comment on “structural and magnetic properties of the new brownmillerite oxides La1−xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3)”, Materials Chemistry and Physics (2016), doi: 10.1016/j.matchemphys.2016.07.055. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Comment on “Structural and magnetic properties of the new brownmillerite oxides La1-xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3)” Josie E. Aucketta*, Chris D. Linga a

RI PT

School of Chemistry F11, The University of Sydney, New South Wales 2006, Australia *Present address: The Bragg Institute, Australian Nuclear Science and Technology Organisation, Lucas Heights, New South Wales 2234, Australia Corresponding author: [email protected]

Keywords: Ceramics; Magnetic materials; Magnetometer; Magnetic properties; Powder diffraction

SC

Abstract

TE D

M AN U

Mohamed et al. recently reported in this Journal on the synthesis and characterisation of a new manganite series, La1-xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3). On the basis of Rietveld refinements against powder x-ray diffraction data, the authors stated that all synthesised members of the series crystallised in the brownmillerite structure with Pnma symmetry. However, the diffraction patterns presented in that paper do not contain key indicators of brownmillerite-like structural distortions, and instead resemble simulated patterns for oxidised perovskite phases. This is consistent with the reported synthesis conditions, which do not include any reaction steps under the strongly reducing conditions normally required to obtain the mixed +2/+3 oxidation state of Mn in the expected brownmillerites. The materials synthesised under the reported conditions were most likely oxidised pseudo-cubic perovskites with δ ≥ 0.5. The authors’ subsequent discussion of the magnetism of La1-xNaxSrMn2O5+δ, which is based upon incorrectly determined coordination environments and oxidation states the Mn ions, is therefore flawed. 1. Crystallography of La1-xNaxSrMn2O5+δ

AC C

EP

The brownmillerite series (La,Sr)2Mn2O5 has been subject to many recent studies of its crystallographic [1-3] and magnetic [4] properties. In a recent paper, Mohamed et al. [5] reported the synthesis of a new series, La1-xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3), assigning it the brownmillerite structure with Pnma symmetry on the basis of x-ray diffraction (XRD) data. The Rietveld refinement diagrams for La0.9 Na0.1SrMn2O5+δ and La0.8 Na0.2SrMn2O5+δ presented by in Section 3 of this paper both contain very large numbers of reflection markers that are not matched by observed reflections in the XRD patterns. Only about 11 peaks with significant intensity are visible in each data set, excluding some very weak features at 2θ ≈ 28, 36 and 47° which are unmatched by any of the >100 markers generated by the brownmillerite model and which probably correspond to a minor impurity phase such as MnO2. As noted by the authors, the 11 observed peaks are rather broad, perhaps due to a slight splitting of the major reflections. Nevertheless, there are far too many reflection markers under the “empty” regions of the pattern for the brownmillerite model used by Mohamed et al. to be justified by the data. Before considering peak splitting, the 11 reflections visible in the XRD patterns can be fully indexed to a small, high-symmetry unit cell such as that corresponding to a typical cubic

ACCEPTED MANUSCRIPT

EP

TE D

M AN U

SC

RI PT

perovskite. Without access to the raw XRD data, we cannot determine whether the apparent peak broadening can be modelled using standard profile parameters and strain factors, but it should be noted that perovskites are also capable of various symmetry-lowering tilts and supercells which cause splitting of the primary reflections [6] and this could reasonably account for the observed broadening. In fact, an XRD pattern simulated for a simple tetragonal perovskite supercell with LaSrMn2O5.5 stoichiometry [Mori] shows an excellent visual match to the data of Mohamed et al. (Figure 1a). The structure of this supercell is shown in Figure 2a.

AC C

Figure 1. XRD patterns (Cu Kα radiation) simulated for various phases using RIETAN [7]. (a) LaSrMn2O5.5, a tetragonal perovskite (ICSD 154546 [3]). (b) Ca2Fe2O5, an archetypal brownmillerite with Pnma symmetry (ICSD 155632 [8]). (c) A modification of the Ca2Fe2O5 structure file with La0.9 Na0.1SrMn2O5 stoichiometry and lattice parameters copied from Table 2 of Mohamed et al. [5].

ACCEPTED MANUSCRIPT (b)

M AN U

SC

RI PT

(a)

Figure 2. (a) Various structures classified as perovskite. Clockwise from top left: cubic perovskite (Pm-3m), a distorted perovskite (I4/mmm), double perovskite (Fm-3m). Unit cell dimensions are shown with black lines. (b) The brownmillerite structure with Pnma symmetry. (Images drawn using VESTA [9])

AC C

EP

TE D

It is important at this point to distinguish clearly between perovskite phases (cubic or distorted) and the brownmillerite structure, as some confusion is evident in the manuscript of Mohamed et al. All perovskite phases ABO3-y (0 ≤ y < 1) have the same average topology, with octahedral BO6 units corner-linked in a three-dimensional cubic network (Figure 2a). This definition includes the ideal cubic perovskite structure as well as its many tilt variations with tetragonal or lower symmetry, and also some supercells caused by ordering of site vacancies or mixed-site cations. By contrast, the brownmillerite structure is defined by a specific ordering of oxygen vacancies corresponding to y = 0.5 stoichiometry, and results in an orthorhombic supercell a ≈ c ≈ √2aP, b ≈ 4aP (aP = cubic perovskite lattice parameter) with equal numbers of BO6 octahedra and BO4 tetrahedra stacked in alternating layers along b (Figure 2b). The substantial distortions of the polyhedra induced by anisotropic vacancy ordering give rise to several distinctive features in the diffraction patterns of brownmillerites, most notably the splitting of the most intense (110)P perovskite reflection into three wellresolved peaks (brownmillerite (200), (002) and (141)). By way of illustration, a simulated XRD pattern generated for a typical brownmillerite with Pnma symmetry, Ca2Fe2O5, is shown in Figure 1b. The absence of this feature in the diffraction patterns of Mohamed et al. shows that if brownmillerite-like vacancy ordering is present in their samples, it has induced far less structural distortion than is usual for such materials. The (020) reflection corresponding to the brownmillerite octahedral/tetrahedral stacking axis is also strongly diagnostic and should be clearly visible in the range 10° < 2θ < 12° for Cu Kα radiation. Unfortunately, the data presented by Mohamed et al. do not include this angular range. However, if the authors were to reinvestigate their samples and find this reflection to be absent, it would indicate almost beyond doubt that long-range brownmillerite vacancy ordering is not present in the La1-xNaxSrMn2O5+δ samples.

ACCEPTED MANUSCRIPT

SC

RI PT

Figure 1c shows an XRD pattern simulated for the Ca2Fe2O5 Pnma brownmillerite structure adjusted to La0.9 Na0.1SrMn2O5 stoichiometry, and with lattice parameters matching those in Table 2 of Mohamed et al. Even though the refined lattice parameters eliminate the splitting of the intense reflection near 33°, many clear and substantial differences exist between this pattern and the XRD data of Mohamed et al., most notably the presence of several simulated peaks of significant intensity which are not observed in the data. The question then arises as to how the authors were able to successfully refine a “brownmillerite” structure against their data. [footnote] A likely explanation lies in the authors’ mention of preferred orientation parameters, which can make entire classes of reflections disappear completely from the calculated XRD profile when allowed to refine to extreme values. It is our opinion that obtaining a good fit to the data of Mohamed et al. using a brownmillerite-type structure model would not have been possible without adopting physically unreasonable preferred orientation parameters, although we note that these parameters were not reported in Table 2 of their paper. 2. Composition of La1-xNaxSrMn2O5+δ

AC C

EP

TE D

M AN U

It is stated in Section 2 of the paper that the samples were quenched in ice water “to prevent the oxidation of Mn3+ to Mn4+”. Elsewhere in the text, the authors describe predicted interactions between Mn2+, Mn3+ and Mn4+ ions supposed to be present in the octahedral and tetrahedral B cation sites of their samples. Neither of these contradictory ideas appears to be based upon any firm evidence, as Mohamed et al. used no experimental methods to verify the Mn oxidation states present in their samples, nor to determine the oxygen stoichiometry parameter δ which should correlate with the sample average Mn oxidation state. Earlier work by the same authors [10] cited an x-ray absorption near-edge structure (XANES) spectroscopic study by Cortes-Gil et al. [4] in support of the unusual triple coexistence of oxidation states in LaSrMn2O5, but that article has not been referenced in the paper under discussion. Nevertheless, Mohamed et al. appear to be strongly influenced by the uncited work of Cortes-Gil et al., as the mixed oxidation states they assign to the “octahedral” and “tetrahedral” Mn sites in their samples are not even consistent with the bond valence sum (BVS) calculations presented in Table 3 of their own paper. In particular, their claim that the octahedral site contains both Mn3+ and Mn4+ ions contradicts the calculated BVS values of >4 which they obtain for all phases. The authors also state that the tetrahedral site contains a mixture of Mn2+ and Mn4+ ions, without appearing to consider more likely combinations such as Mn2+ and Mn3+. Because the BVS is an empirical quantity determined by the distances between atoms, its reliability depends entirely upon the accuracy of the atomic positions in the structure refined against XRD data. If the wrong structure has been refined, as we have argued in the previous section, then the BVS values in Table 3 of the paper are meaningless. As we have previously noted [11], actively reducing conditions are required to synthesise brownmillerite-type LaSrMn2O5 phases [1, 2]. Given that fully oxidised perovskite phases (La,Sr)MnO3 can be obtained by high-temperature synthesis in air [2], we consider it highly unlikely that simple quenching of air-annealed La1-xNaxSrMn2O5+δ samples was sufficient to yield the reduced brownmillerite structure. In light of the simulated diffraction patterns in Figure 1, it is far more likely that the obtained phases are perovskites with 0.5 ≤ δ ≤ 1. This view is also supported by the fact that the unit cell volume determined by Mohamed et al. for the synthesised x = 0 phase (LaSrMn2O5+δ), when divided by (√2×√2×4) to obtain the

ACCEPTED MANUSCRIPT primitive perovskite volume of 57.68 Å3, lies remarkably close to the equivalent volume reported by Chmaissem et al. [12] for the ferromagnetic distorted perovskite La0.5Sr0.5MnO3 (V/(√2×√2×2) = 57.51 Å3). 3. Magnetic analysis of La1-xNaxSrMn2O5+δ

M AN U

SC

RI PT

As demonstrated above, the coordination environments of Mn in La1-xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3) have been incorrectly determined by Mohamed et al. due to the fitting of an inappropriate structural model to the XRD data, and the oxidation states of Mn are uncertain due to the lack of experimental evidence and the absence of reducing conditions used during synthesis. Therefore, the discussions of magnetic exchange interactions in La1-xNaxSrMn2O5+δ presented in that paper are invalid. It is also worth noting here that ferromagnetic brownmillerites are unprecedented in the literature, with the exception of a small number of G-type antiferromagnetic phases that display weak ferromagnetic canting [13]. Taking all of these factors into account, the magnetisation data of Mohamed et al. should be reinterpreted in the context of the correct crystallographic structure, which is probably a low-symmetry perovskite, along with a rigorous experimental determination of the oxygen content of the samples.

References

4. 5. 6. 7. 8. 9. 10. 11.

TE D

3.

EP

2.

Casey, P.S., D. Barker, and M.A. Hayward, Charge and structural ordering in the brownmillerite phases: La1-xSrxMnO2.5 (0.2 < x < 0.4). Journal of Solid State Chemistry, 2006. 179(5): p. 1375-1382. Parsons, T.G., et al., Synthesis and Structural Characterization of La(1-x)A(x)MnO(2.5) (A = Ba, Sr, Ca) Phases: Mapping the Variants of the Brownmillerite Structure. Chemistry of Materials, 2009. 21(22): p. 5527-5538. Mori, T. and N. Kamegashira, The crystal structure of an oxygen-deficient manganite perovskite LaxSr1−xMnO(5+x)/2 (0 ≤ x ≤ 0.5). Journal of Alloys and Compounds, 2006. 408– 412: p. 1210-1213. Cortes-Gil, R., et al., Magnetic Structure and Electronic Study of Complex Oxygen-Deficient Manganites. Chemistry-a European Journal, 2008. 14(29): p. 9038-9045. Mohamed, M., et al., Structural and magnetic properties of the new brownmillerite oxides La1−xNaxSrMn2O5+δ (0.1 ≤ x ≤ 0.3). Materials Chemistry and Physics, 2015. 166: p. 49-56. Howard, C.J. and H.T. Stokes, Group-theoretical analysis of octahedral tilting in perovskites. Acta Crystallographica Section B-Structural Science, 1998. 54: p. 782-789. Izumi, F. and T. Ikeda, A Rietveld-Analysis Programm RIETAN-98 and its Applications to Zeolites. Mater. Sci. Forum, 2000. 321-324: p. 198-205. Shaula, A.L., et al., Ionic conductivity of brownmillerite-type calcium ferrite under oxidizing conditions. Solid State Ionics, 2006. 177(33-34): p. 2923-2930. Momma, K. and F. Izuma, VESTA: a three-dimensional visualization system for electronic and structural analysis. J. Appl. Crystallogr., 2008. 41: p. 653-658. Mohamed, M., et al., Structural and Mossbauer study of the Brownmillerite oxides LaSrMn2xFexO5 (0 <= x <= 0.5). Journal of Alloys and Compounds, 2013. 581: p. 378-384. Auckett, J.E. and C.D. Ling, Comment on “Structural and Mössbauer study of the brownmillerite oxides LaSrMn2−xFexO5 (0 ⩽ x ⩽ 0.5)”. Journal of Alloys and Compounds, 2014. 610: p. 212-213.

AC C

1.

ACCEPTED MANUSCRIPT 12. 13.

Chmaissem, O., et al., Structural and magnetic phase diagrams of La1-xSrxMnO3 and Pr1ySryMnO3. Physical Review B, 2003. 67(9). Auckett, J.E., et al., Neutron Laue diffraction study of the complex low-temperature magnetic behaviour of brownmillerite-type Ca2Fe2O5. Journal of Applied Crystallography, 2015. 48(1): p. 273-279.

AC C

EP

TE D

M AN U

SC

RI PT

[footnote] No atomic parameters are presented in the paper, so we are unable to confirm that the refined structures actually resembled brownmillerites when the Rietveld fits shown in Figure 3 of the paper were achieved. However, it can be assumed from the lattice parameters reported in Table 2 of the paper that brownmillerite-like structures were used in the refinements, rather than the closely related A3B3O8-type structure (a ≈ c ≈ √2aP, b ≈ 3aP) described incorrectly as “brownmillerite” in the first paragraph of the Introduction, or any other low-symmetry perovskite variant.

ACCEPTED MANUSCRIPT

RI PT SC M AN U TE D EP

• •

XRD data for La1-xNaxSrMn2O5+δ samples of Mohamed et al. are inconsistent with the brownmillerite structure Evidence clearly supports distorted perovskite structures for the reported phases Samples are probably over-oxidised due to inappropriate synthesis conditions

AC C

•